Assignment 1: Reproduce the number

Take a number somebody published and get it back.

Choose a published FID, rebuild enough of the pipeline to compute it yourself, and report what you get beside what they got. Then account for the difference. Almost nobody closes it, and the mark is in the account rather than in the size of the gap.

Due end of week 5, which is the week that explains why the standard deviations you computed did not warn you. Deadlines, extensions and the use of generative tools are course-wide and live on the policies page.

Before you start#

1. Individual work. 100 points, worth 20 per cent of the course.
2. Three parts. Part A is 30 points, Part B 50, Part C 20. The marking criteria on this page are those three parts.
3. Generative tools are governed by the policies page. Read it before you start rather than before you submit.
4. Submit one PDF containing Parts A to C, at most four pages, and one archive holding everything needed to re-run Part A on a clean machine.

Part A, the reproduction [30]#

Choose a published FID from a paper of your choice. Cite the paper, the table and the page. Rebuild enough of the pipeline to compute it. Report your number beside theirs, with N and implementation stated for both.

Match what you can: the same reference set if it is obtainable, the same N if the paper states one, the same instrument weights if you can find them. Where you cannot match, that is Part B’s material rather than a failure.

Part B, the account [50]#

Every difference between your setup and theirs, named. Where the paper does not say, state what you assumed and why.

This part is marked on completeness and honesty, not on the size of the gap. A submission that closes the gap and cannot say how is worth less than one that misses by a wide margin and can account for every step of it. The strongest submissions are the ones where the assumptions are visible: what could not be determined from the paper, what was assumed in its place, and roughly what each assumption could be worth.

Part C, the comparison [20]#

Was your comparison N-honest? If it was not, say what it would take to make it so, and what that would cost.

In person#

Ten minutes, week 6, at the codewalk, which carries the day, the rooms and the booking. You bring the PDF and the archive. We pick one difference from Part B and ask you to defend it. Bring your ID.